1.4 Question Style and Score Report Thinking
Key Takeaways
- DP-700 mixes single-answer multiple choice, multiple-response, drag-and-drop ordering, and one or more case studies.
- Case studies bundle several questions around a shared scenario and may lock answers once you advance past a section.
- Unscored pretest items appear but are unmarked, so treat every question as if it counts.
- The score report breaks performance down by skill area, which tells you exactly where to remediate.
- Read the stem for the task verb and the decisive cue before comparing options.
The question formats you will see
DP-700 is not one format. Across the roughly 50 items you should expect a mix that tests both recognition and applied judgment:
| Format | What it asks | Strategy |
|---|---|---|
| Single-answer multiple choice | Pick the one best action or setting | Eliminate options that ignore the decisive cue |
| Multiple-response (choose two/three) | Select every correct option | A partially correct set usually earns no credit |
| Drag-and-drop / ordering | Sequence steps or match items to roles | Build the workflow order mentally first |
| Hot area / build list | Configure or assemble a solution | Watch for distractor steps that look plausible |
| Case study | Several questions on one shared scenario | Read the exhibits before answering |
Case studies are the format candidates most often mishandle. A case study presents a business scenario with multiple tabs or exhibits (requirements, existing environment, technical constraints) and then asks a cluster of questions against it. In Microsoft exams, once you finish a case-study section and move on, you typically cannot return to change those answers, so read all the exhibits carefully before committing. Budget extra time: a single case study can consume 15-20 minutes.
Pretest items and timing
Microsoft seeds unscored pretest items into the exam to trial future questions. They are not marked, so do not waste effort trying to spot them; answer every item with the same process. With about 50 questions in 100 minutes you have roughly two minutes each, but case studies skew that. Flag-and-return is available for standalone items, so move past a hard standalone question rather than burning eight minutes on it.
A reliable reading method
Read the stem before the options, in this order: identify the task verb (configure, optimize, ingest, secure, troubleshoot), identify the skill area it belongs to, then find the decisive cue. The cue is the one detail that separates two plausible answers, for example "without copying the data" (shortcut, not mirroring), "data is missing after deployment" (refresh the semantic model), or "resume reliably after a restart" (checkpointed streaming). Only after naming the cue should you compare options, because a familiar Fabric term in a wrong answer is a classic distractor.
- Read the task verb.
- Name the skill area.
- Underline the decisive cue.
- Eliminate options that violate a Fabric default or ignore the cue.
- Choose the most specific, complete answer.
- Flag and log the miss by area and cause.
Reading the score report
Whether you pass or fail, the score report breaks your performance down by the three skill areas. This is the most actionable feedback Microsoft gives you. If you fail, the bar chart shows which area dragged you down, so you can target remediation rather than restudying everything. In practice, mirror this on every practice set: tag each miss with its skill area and a cause category (content gap, misread cue, wrong sequence, wrong tool choice, or changed a right answer to wrong). After a few sessions, the pattern of misses by area tells you exactly where the remaining points are.
High-yield decision cues to recognize
Many DP-700 items hinge on a single phrase. Training yourself to react to these recurring cues turns slow reasoning into fast pattern-matching:
| Cue in the stem | What it points to |
|---|---|
| "without copying the data" | OneLake shortcut, not mirroring or a copy pipeline |
| "continuously synchronized / low latency replica" | Mirroring of the supported source |
| "data is missing after deployment" | Refresh the semantic model; pipelines copy metadata, not data |
| "resume reliably after a restart" | Checkpointed streaming into Delta |
| "read-only access to view content" | Viewer workspace role |
| "custom Python libraries, distributed processing" | Spark notebook |
| "start when a file arrives" | Event-based trigger, not a schedule |
| "reduce repeated cross-cloud reads" | OneLake shortcut caching |
Managing the clock and your nerves
With roughly 50 items and case studies that can each eat 15-20 minutes, pace yourself: aim to finish standalone questions in about 90 seconds so case studies have room. Use flag-and-return on standalone items you are unsure about, but remember case-study answers usually lock when you advance. If you blank on an item, eliminate the options that violate a known Fabric default, then commit and flag it rather than freezing. Changing answers late is risky: data on test-taking shows a confident first read is often correct, so only revise a flagged item when you have a concrete reason, such as a cue you missed on the first pass.
Distractor patterns to expect
Microsoft's distractors are engineered, not random, and recognizing the pattern speeds elimination. A common pattern is the right tool for the wrong layer: an answer names a real Fabric item (a Dataflow Gen2, a notebook, a pipeline) but applies it to a task another item handles better. Another is the outdated default: an option states a behavior that was once true or that sounds intuitive but contradicts a current default, such as claiming deployment pipelines move data.
A third is the over-broad security grant: an option fixes access by giving Admin or tenant-wide rights when a narrower role or a single OneLake data-access role is correct. A fourth is the plausible-but-incomplete sequence in ordering items, where one step is missing or two are swapped. When two options survive your first cut, prefer the one that is most specific to the stem's cue and leaves the cleanest, least-privileged, most auditable result.
Why should a DP-700 candidate read all exhibits of a case study before answering its first question?
How should a candidate treat the unscored pretest items that Microsoft seeds into DP-700?
What is the most useful piece of feedback on the DP-700 score report?
A multiple-response DP-700 item asks you to "select two" correct configuration steps, and you are confident about only one. What is the best approach?